Transform requirements into formal specifications using EARS methodology
Version: v3.12.0 | Archetype: Spec-Engineer | Skills: 2 specialized + 15 universal
The Spec-Engineer archetype brings requirements engineering rigor to specification work. Unlike informal documentation, spec-engineer agents provide:
- Spec validation — Check specifications for ambiguity, completeness, and testability
- Requirement generation — Transform informal needs into formal EARS-pattern requirements
- Traceability — Connect requirements to implementation with verifiable links
For evaluators: If you need an AI that can formalize requirements into aerospace-grade specifications, the Spec-Engineer archetype brings EARS methodology to your requirements process.
Domain knowledge that compounds: Spec-engineer agents build persistent understanding of your requirements landscape — specification patterns, domain terminology, and validation criteria. Unlike tools that start fresh each session, your agent accumulates specification context that makes each requirement more precise and each validation more thorough.
Spec-Engineer agents come with 2 archetype-specific skills plus the universal AGET skills.
| Skill | Description |
|---|---|
| aget-validate-spec | Validate specifications for ambiguity, completeness, and testability. Identifies gaps and conflicts. |
| aget-generate-requirement | Generate formal requirements using EARS patterns (ubiquitous, event-driven, state-driven, optional-feature). |
All AGET agents include session management, knowledge capture, and health monitoring:
aget-wake-up/aget-wind-down— Session lifecycleaget-create-project/aget-review-project— Project managementaget-record-lesson/aget-record-observation— Learning captureaget-check-health/aget-check-kb/aget-check-evolution— Health monitoringaget-propose-skill/aget-create-skill— Skill developmentaget-save-state/aget-file-issue— State and issue management
Spec-Engineer agents use a formal vocabulary of 7 concepts organized into 2 clusters:
| Cluster | Concepts |
|---|---|
| Requirements | Requirement, Capability, Constraint |
| EARS Patterns | EARS_Pattern, Trigger, Response, Rationale |
This vocabulary enables precise communication about specifications.
See: ontology/ONTOLOGY_spec_engineer.yaml
# 1. Clone the template
git clone https://github.com/aget-framework/template-spec-engineer-aget.git my-spec-engineer-agent
cd my-spec-engineer-agent
# 2. Configure identity
# Edit .aget/version.json:
# "agent_name": "my-spec-engineer-agent"
# "domain": "your-domain"
# 3. Verify setup
python3 -m pytest tests/ -v
# Expected: All tests passing# In Claude Code CLI
/aget-validate-spec # Check a specification
/aget-generate-requirement # Create EARS requirement| Aspect | Informal Docs | Spec-Engineer Agent |
|---|---|---|
| Requirements | Ambiguous prose | EARS-pattern formal statements |
| Validation | Human intuition | Systematic ambiguity detection |
| Testability | Hoped for | Verified at specification time |
| Patterns | Ad-hoc | Ubiquitous, event-driven, state-driven |
| Domain memory | Starts fresh each session | Accumulates specification expertise over time |
| Directory | Purpose | Owner |
|---|---|---|
.claude/skills/ |
Slash command definitions | Framework + Agent |
.claude/agents/ |
Subagent definitions | Agent |
.claude/rules/ |
Path-scoped context rules | Agent |
Skills are provided by the template. Agents and rules directories are scaffolded for your customization.
| Attribute | Value |
|---|---|
| Framework | AGET v3.12.0 |
| Archetype | Spec-Engineer |
| Skills | 17 total (2 archetype + 15 universal) |
| Ontology | 7 concepts, 2 clusters |
| License | Apache 2.0 |
- AGET Framework — Core framework documentation
- Archetype Guide — All 12 archetypes explained
- Getting Started — Full onboarding guide
| Archetype | Best For |
|---|---|
| Architect | System design driving specs |
| Developer | Implementing specs |
| Reviewer | Validating spec compliance |
AGET Framework | Apache 2.0 | Issues